Anomaly detection UI case study · 2024

Anomaly Detection UI

Anomaly Detection UI

A concept for a marketing analytics platform that turns unusual performance shifts into clear, actionable insights.

Role

Product Designer

Scope

UX/UI Design · Mobile Product Design · Design System

Focus

Anomaly detection and collaboration workflows

The goal was to help users build healthier routines without making them feel pressured by another complicated self-improvement app.

THE CHALLENGE

Marketing teams often work across several campaigns, channels, and metrics at once. When performance suddenly changes, finding the issue is only the first step.

The real challenge is understanding what changed, deciding whether it matters, and getting the right person involved before it affects performance further.

THE IDEA

I designed the experience around one simple loop:

Stop it

Understand it

Act on it

Instead of surfacing a long list of raw alerts, the product brings important changes forward through severity, clear visual signals, and contextual explanations.

USER FLOW

Anomaly summary panel: metric, deviation %, detected date, etc.

Action panel: Mark as Resolved, Assign to Teammate, Share Insight

Actual vs Expected line chart and insight box with contextual copy

DESIGNING FOR

Marketing teams who need to understand what changed before it becomes a bigger problem.

This experience is designed for people who monitor campaign performance across several platforms. They do not need more alerts. They need a clear signal, enough context to understand it, and an easy way to act with their team.

Emily Cartes, 36

Senior Digital Marketing Manager · London, UK

“I need to know what is worth looking into, why it changed, and who should take it from there.”

What she needs

A clear view of the most important anomalies

Context behind a sudden performance shift

A simple way to assign, resolve, or share an insight

Less switching between dashboards and team tools

OUTCOME

The result is a clearer workflow for teams working with campaign data. It helps users move from “something changed” to “here is what to do next” with less effort.